Cannot use bindTools and withStructuredOutput together

[Obada-sh]

Checked other resources

• I added a very descriptive title to this issue.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).
• I posted a self-contained, minimal, reproducible example. A maintainer can copy it and run it AS IS.

Example Code

import { Injectable } from "@nestjs/common";
import { ChatOpenAI } from "@langchain/openai";
import { NewsSchema, News } from "./news.schema";

@Injectable()
export class LlmService {
  private readonly llm = new ChatOpenAI({
    model: "gpt-4o-mini",
    temperature: 0.2,
    openAIApiKey: "myKey",
  })
    .bindTools([{ type: "web_search_preview" }])
    .withStructuredOutput(NewsSchema, {
      strict: true,
    });

  async positiveNewsToday(): Promise<News> {
    return this.llm.invoke("Give me one positive news story from today");
  }
}

Error Message and Stack Trace (if applicable)

No response

Description

I’m trying to build an agent that both has tools bound (e.g. a web‐search preview tool) and emits structured output via my own schema. However, calling .bindTools([...]) before .withStructuredOutput(...) causes an error and prevents me from using structured output at all.

System Info

System Info (Node.js / NestJS)

node -v : v22.12.0
npm -v : 10.9.0
npm ls @nestjs/core : teest@0.0.1 E:\Work\teest
├── @nestjs/core@11.1.3
├─┬ @nestjs/platform-express@11.1.3
│ └── @nestjs/core@11.1.3 deduped
└─┬ @nestjs/testing@11.1.3
└── @nestjs/core@11.1.3

[hntrl]

@Obada-sh Thanks for the issue!

This is a known limitation with the current Runnable interface. It was first designed before the introduction of structured outputs/ tools, so there were some zero-day incompatibilities with the current methodology that model providers employ.

We have it slated to address this for the next minor version/ first major version of langchain. I'll keep this issue up just so, but it might take a while for a fix to come for this.